1
Reconfigurable Intelligent Surface-aided M -ary
FM-DCSK System: a New Design for Noncoherent
Chaos-based Communication
Huan Ma, Yi Fang, Member, IEEE, Pingping Chen, Member, IEEE, Yonghui Li, Fellow, IEEE
Abstract—In this paper, we propose two reconfigurable intelli-
arXiv:2206.07997v1 [eess.SP] 16 Jun 2022
gent surface-aided M -ary frequency-modulated differential chaos
shift keying (RIS-M -FM-DCSK) schemes. In scheme I, the RIS is
regarded as a transmitter at the source to incorporate the M -ary
phase-shift-keying (M -PSK) symbols into the FM chaotic signal
and to reflect the resultant M -ary FM chaotic signal toward the
destination. The information bits of the source are carried by
both the positive/negative state of the FM chaotic signal and the
M -PSK symbols. In scheme II, the RIS is treated as a relay
so that both the source and relay can simultaneously transmit
their information bits to the destination. The information bits of
the source and relay are carried by the positive/negative state
of the FM chaotic signal and M -PSK symbols generated by
the RIS, respectively. The proposed RIS-M -FM-DCSK system
has an attractive advantage that it does not require channel
state information for detection, thus avoiding complex channel
estimation. Moreover, we derive the theoretical expressions for
bit error rates (BERs) of the proposed RIS-M -FM-DCSK system
with both scheme I and scheme II over multipath Rayleigh fading
channels. Simulations results not only verify the accuracy of the
theoretical derivations, but also demonstrate the superiority of
the proposed system. The proposed RIS-M -FM-DCSK system is
a promising low-cost, low-power, and high-reliability alternative
for wireless communication networks.
Index terms— Reconfigurable intelligent surface (RIS), dif-
ferential chaos shift keying (DCSK), channel state information,
multipath Rayleigh fading channel.
I. I NTRODUCTION
Chaos-based communication is considered as a potential
alternative to the conventional direct sequence-spread spec-
trum communication [1], [2]. Thanks to natural characteris-
tics of chaotic signal, chaos-based communication benefits
from various advantages such as excellent anti-intercept, auto-
correlation, non-periodic, and wideband properties [3]. In the
past three decades, differential chaos shift keying (DCSK),
which is regarded as the most popular chaos-based modulation,
has owned considerable development. DCSK only needs a
non-coherent demodulator to avoid the chaos synchronization,
which is very desired for practical implementation in low-
power and low-cost wireless-communication applications.
The theoretical bit-error-rate (BER) performance of DCSK
system in different types of wireless channels has been inves-
H. Ma and Y. Fang are with the School of Information Engineering,
Guangdong University of Technology, Guangzhou 510006, China, (email: mh-
zs@163.com; fangyi@gdut.edu.cn).
P. Chen is with the Department of Electronic Information, Fuzhou Univer-
sity, Fuzhou 350116, China (e-mail: ppchen.xm@gmail.com).
Y. Li is with the School of Electrical and Information Engineer-
ing, The University of Sydney, Sydney, NSW 2006, Australia (e-mail:
yonghui.li@sydney.edu.au).
tigated in [4], [5]. With the above theoretical advances, the
DCSK system has been further designed to adapt to different
scenarios, such as magnetic coupled human body communica-
tion [6], power line communication [7], underwater acoustic
communication [8], [9], ambient backscatter communication
[10] and vehicle-to-vehicle communication [11]. To improve
the transmission reliability of DCSK system, some powerful
error correction codes, such as low-density parity-check codes
[12], [13], have been exploited to construct highly robust
coded modulation systems [14].
In recent years, various variants of the DCSK system have
been presented. To avoid the drawback of the non-constant en-
ergy in the conventional DCSK system, a frequency-modulated
(FM) DCSK system has been devised in [15]. Afterwards,
some ultra-wideband systems based on FM-DCSK modulation
have been proposed [16]–[18]. Although the introduction of
FM can improve the performance of traditional DCSK system,
it data rate and energy efficiency are low because half of
the chaotic chips are used to transmit the reference signal.
Therefore, designing enhanced DCSK systems with higher
data rate and energy efficiency has become an urgent issue.
In particular, a code-shifted DCSK system with a delay-line-
free receiver has been proposed in [19], where the reference
and information-bearing signals are sent in the same time
slot. An improved code-shifted DCSK system with reference
diversity has been designed in [20]. Thanks to the reference
diversity, this system has better BER performance than the
conventional code-shifted DCSK system over fading channels.
To make full use of frequency domain resources, some multi-
carrier (MC) DCSK systems have been proposed [21], [22].
These MC-DCSK systems not only have the advantage of the
delay-line-free structure, but also have higher data rate than
the conventional DCSK system. To further improve the data
rate and efficiency of the DCSK system, the joint design of
the index modulation (IM) technique and the DCSK system
has been proposed recently. Besides, a code-index-modulation
(CIM) DCSK has been developed in [23] and optimized in
[24]. In the CIM-DCSK system, the row index of the selected
Walsh code sequence in the Walsh matrix is used to convey
the information bits. In [25], [26], more than one transmission
entities are used for IM. A joint carrier-code IM-aided M -
ary DCSK system was proposed in [25]. In such a system, the
carrier and the row index of the selected Walsh code sequence
are used to carry extra information bits. In [26], a joint
time-frequency-IM-aided system has been proposed, where all
subcarriers and time slots are used to carry information bits.
Although various excellent DCSK variants have been pro-
2
posed, how to maintain the reliability of the DCSK system
in transmission scenarios with relatively long distance or
some obstacles still encounters great challenges. To achieve
better reliability for relatively long-distance communications,
an analog space-time block coded (STBC) DCSK system
was devised in [27]. Compared with the single-antenna
DCSK system, the analog STBC-DCSK system has a great
improvement in terms of BER performance over multi-
path fading channels. An orthogonal multi-codes multiple-
input-multiple-output (OMC-MIMO) MIMO-DCSK system
was proposed in [28] to further improve the BER per-
formance. To boost the data rate of the multiple-antenna
DCSK system, a multiple-input-single-output code-index-
modulation-aided multiple-carrier M -ary DCSK (MISO-CIM-
MC-M DCSK) system was proposed in [29], which inherits
the merits of DCSK modulation, multiple-antenna technique,
MC technique, and IM technique. To overcome the adverse
effects of obstacles in transmission, some cooperative DCSK
systems have been proposed, such as the amplify-and-forward
relaying DCSK system [30] and two-way relay network-coded
DCSK system [31]. However, the relays of these systems are
active, which is not conducive to the deployment in practical
applications. To resolve this problem, a simultaneous wireless
information and power transfer (SWIPT) cooperative DCSK
system with passive relay has been proposed in [32]. Further-
more, a buffer-aided DCSK-SWIPT system was developed in
[33]. In this system, the relay works by harvesting the energy
of the received signal. Besides, two link-selection protocols
have been designed for the buffer-aided DCSK-SWIPT system.
As a passive-communication-aided device, reconfigurable
intelligent surface (RIS) is a promising technology [34], [35].
The RIS is a plane consisting of a large number of passive
reflective elements, each of which can control the amplitude
and/or phase of the incident signal [36]. Based on these char-
acteristics, the RIS has the distinctive capability to transform
uncontrollable and bad channel condition into controllable and
good channel condition when the channel state information
(CSI) is available at the RIS [37]. On the contrary, when
the CSI is unavailable at the RIS, though the RIS cannot
maximize the received SNR according to the channel phases,
it can improve the received SNR using the passive reflective
elements [38], [39]. Unlike multi-antenna and relay solutions,
the RIS does not require multiple radio frequency (RF) chains,
hence enabling the advantages of low power and low cost [40].
Motivated by the above advantages of RIS, we conceive
an RIS-aided M -ary FM-DCSK system (i.e., RIS-M -FM-
DCSK system) in this paper, which can realize high-reliability
transmissions over multipath Rayleigh channels. In addition,
the RIS does not require any CSI in the proposed RIS-M -FM-
DCSK system. The contributions of this paper are summarized
as follows:
1) We develop two different transmission schemes, referred
to as scheme I and scheme II, to improve the reliability
for the proposed RIS-M -FM-DCSK system. Scheme I
is suitable for the communication scenario (CS) that the
direct link exists between the source and the destination.
In this scheme, the RIS is considered as a transmitter at
the source. Moreover, the information bits are not only
carried by the positive/negative state of the FM chaotic
signal, but also carried by the M -ary phase-shift-keying
(M -PSK) symbols obtained by the phase adjustment of
the RIS. Scheme II is suitable for the CS that the direct
link between the source and the destination is blocked by
an obstacle. In this scheme, the RIS is treated as a relay to
realize the information transmissions simultaneously from
the source and relay to the destination. Moreover, the
information bits of the source and relay are carried by the
positive/negative state of the FM chaotic signal and M -
PSK symbols generated by the RIS, respectively.
2) We carefully analyze the theoretical BER performance of
the two proposed transmission schemes in the RIS-M -FM-
DCSK system over multipath Rayleigh fading channels.
3) We carry out various simulations to verify the accuracy
of the proposed BER expressions. Furthermore, we illus-
trate that the proposed RIS-M -FM-DCSK system achieves
significant performance gains over the existing DCSK
counterparts.
The remainder of this paper is organized as follows. The
system model of the proposed RIS-M -FM-DCSK system is
presented in section II. Section III carries out the theoretical
BER performance analysis of the proposed system. Section IV
presents the simulation results and discussions. Finally, Sec-
tion V draw the conclusion.
II. S YSTEM M ODEL
In this section, we describe system model of the proposed
RIS-M -FM-DCSK modulation. We consider two CSs: i) the
direct link exists between the source and the destination; ii) the
direct link between the source and the destination is blocked by
an obstacle. In the CS-i, we treat the RIS as a transmitter. In the
CS-ii, we treat the RIS as a relay.1 Based on the characteristics
of RIS, we separately design two transmission schemes, i.e.,
scheme I and scheme II, for the above two CSs. The block
diagrams for the two transmission schemes are illustrated in
Fig. 1.
A. RIS-M-FM-DCSK System: Scheme I
Fig. 1(a) shows the system model for the proposed RIS-
M -FM-DCSK with scheme I. In this scheme, the RIS is
very close to the transmit antenna, hence we assume that
the transmission between the RIS and transmit antenna is
not affect by fading. As such, the RIS can be treated as a
transmitter. Referring to Fig. 1(a), the information bit stream
a = [b, c] of the source is divided into two parts, i.e., the
bit b and the bit stream c = [c1 , . . . , c∂ , . . . , cn ], where the
dimension of c∂ is 1×log2 M and 1 ≤ ∂ ≤ n. Then, the bit b is
converted into a symbol d = 2b − 1, while the bit sub-streams
c1 , . . . , cn are converted to symbols s1 , . . . , sn , respectively,
by using an M -ary bit-to-symbol converter. The number of the
information bits carried by each transmitted symbol in scheme
I equals ΞI,tot = 1 + nlog2 M . To generate a signal that can
carry these ΞI,tot information bits, the length-β chaotic signal
1 Since RIS is treated as a relay in scheme II, “relay” may be used to
represent “RIS” in scheme II from this point onwards.
 3

5,6
6RXUFH 5,6
FRQWUROOHU
'HVWLQDWLRQ
exp M1 ,},exp Mn
s1,est.,},sn,est.
%LWVWRV\PEROV M-36. E
036. 6\PEROWR F1,est.,!, Fn,est.
FRQYHUWHU s1,},sn PRGXODWLRQ As,w ¦Z
i 1
i
˜ ZwEi  Z n1 Ei ˜ Z n1 EwEi
GHWHFWLRQ ELW

,QIRUPDWLRQ
D F ª¬F1,F2,!,Fn º¼ dest.  o 
best.
ELWVWUHDP 0DWUL[= °­
n1 E

Ad Real ® ¦ Zi ˜ Z n1 Ei ¾

°½ 'HWHFWLRQRI %LWV D est.
 o  d °¯ i 1 ¿°
V\PEROdd  o  FRPELQDWLRQ
b  o 

N [k1,!,kE ] 5HSOLFDV 5DGLR

&KDRWLF
)0PRGXODWRU 2(n+1) IUHTXHQF\
JHQHUDWRU
WLPHV JHQHUDWRU

(a)
5HOD\
5,6
s, } , sn
%LWWRV\PERO M-36.
FRQYHUWHU PRGXODWLRQ
exp M1 ,},exp Mn
5,6
F ª¬F 1, F 2 ,!, F n º¼ FRQWUROOHU

d 'HVWLQDWLRQ F 1,est., !, F n ,est.

b
6RXUFH  o  E s1,est., }, sn,est.
 036. 6\PEROWR
 o  As,w ¦Z ˜ Z E
i1
i w i
Z n1 Ei ˜ Z n1 EwEi
GHWHFWLRQ ELW
N [k1,!,kE ] 5HSOLFDV 5DGLR
&KDRWLF
)0PRGXODWRU 2(n+1) IUHTXHQF\
JHQHUDWRU 
'HWHFWLRQRI d est.  o 
best.
WLPHV ­ n1 E ½°
JHQHUDWRU   Real ®° Z ˜ Z
=
0DWUL[Z A ¦ i n1 Ei ¾
d
¯° i 1 ¿° V\PEROd  o 
2EVWDFOH

(b)

Fig. 1. Block diagrams of (a) scheme I and (b) scheme II in the proposed RIS-M -FM-DCSK system.

x = [x1 , x2 , . . . , xβ ] is first modulated by using FM modulator equals the negative of signal Rd . Besides, the signals k (resp.
to generate FM chaotic signal k = [k1 , k2 , . . . , kβ ], where d · k) and exp (jϕ∂ ) k (resp. d · exp (jϕ∂ ) k) in Eq. (2) can
logistic map is employed by chaotic generator and the chip be regarded as reference part and information-bearing part in
duration is defined as Tc . Then, the chaotic signal k is repeated the transmission of symbol s∂ (1 ≤ ∂ ≤ n), respectively. In
2 (n + 1) times to modulate the symbols d and s1 , . . . , sn . this sense, the spreading factor SF of scheme I is defined
More specifically, the symbols s1 , . . . , sn are modulated via as SF = 2 (n + 1) β. Hence, the data rate of scheme I is
Ξ
an M -PSK modulator, thus generating the phases ϕ1 , . . . , ϕn , RI = SFI,tot
·Tc .
where ϕ∂ ∈ (0, 2π] (1 ≤ ∂ ≤ n). Meanwhile, the symbol d is At the destination, the received baseband signal can be
conveyed by the discrete baseband signal TRF , which is sent written as
from the transmit antenna to the RIS. In particular, TRF can
be expressed as
   q Lsd X
X N
−ε
rℑ = Dsd αsd,l,ρsd Tel,l,ℑ−τl,ρsd + wℑ
TRF = k, k, . . . k, d k, k, . . . , k . (1) ρsd =1 l=1
| {z } | {z }
n+1 items n+1 items q Lsd
X
−ε
= Dsd Λsd,ρsd Tel,1,ℑ−τρsd + wℑ , (3)
The RIS reflects the incident signal with phase-related reflec-
√ ρsd =1
tion coefficients exp (jϕ1 ) , . . . , exp (jϕn ), where j = −1.
Therefore, for the l-th RIS reflecting element, the reflected
baseband signal can be expressed as where Dsd is the distance between the RIS and destination, ε is
 the path loss coefficient [41], Lsd is the number of the paths
 between the RIS and destination, αsd,l,ρsd and τl,ρsd denote
Tel,l = k, exp (jϕ1 ) k, . . . , exp (jϕn ) k, channel coefficient and delay of the ρsd -th (1 ≤ ρsd ≤ Lsd )
| {z }
Rd  path between the l-th RIS reflecting element and destina-
(2) tion, respectively, 1 ≤ ℑ ≤ 2 (n + 1) β, and Λsd,ρsd =
PN
l=1 αsd,l,ρsd . In Eq. (3), wℑ ∼ CN (0, N0 ) is complex addi-

d · k, d · exp (jϕ1 ) k, . . . , d · exp (jϕn ) k
| {z } tive white Gaussian noise (AWGN), and CN (φ1 , φ2 ) denotes
Id
  the circularly symmetric complex Gaussian distribution with
= Tel,l,1 , . . . , Tel,l,ℑ , . . . , Tel,l,2(n+1)β ,
mean of φ1 and variance of φ2 . In this paper, we assume
where 1 ≤ l ≤ N , N is the number of the RIS reflecting that the path delays between each RIS reflecting element and
elements, the signals Rd and Id can be regarded as the destination are the same [28], [42]–[44], i.e., τl1 ,ρsd = τl2 ,ρsd
reference part and information-bearing part in the transmission (l1 = 1, . . . , N , l2 = 1, . . . , N , and l1 6= l2 ). Hence, τl,ρsd
of symbol d, respectively. The symbol d is carried by the can be simplified to τρsd . Subsequently,  the received baseband

state of signal Id . To be specific, the signal Id equals the signals are stored in a matrix Z = Z1 , Z2 , . . . , Z2(n+1)β . To
positive of signal Rd if d = 1; otherwise (i.e., d = −1), it perform the demodulation for the symbol d of the source, the
4

decision metric is first formulated as source to the destination is ΞII,s = 1, while the number of
 
(n+1)β  the information bits sent from the relay to the destination is
X
Ad = Real ∗
Zi · Z(n+1)β+i ΞII,r = nlog2 M .

i=1
 Next, at the source, the symbol d˙ is conveyed by the discrete
( n+1 β Lsd
baseband signal ṪRF , which is emitted from the antenna
X X q X
≈ Real −ε
Dsd Λsd,ρsd exp (jϕi1 −1 ) ki2 towards the RIS. Specifically, ṪRF is expressed as
  
i1 =1 i2 =1 ρsd =1
q Lsd
X ṪRF = k, k, . . . k, ḋ k, k, . . . , k . (7)

−ε
+w(i1 −1)β+i2 × d· Dsd Λsd,ρsd exp (jϕi1 −1 ) ki2 | {z } | {z }
n+1 items n+1 items
ρsd =1

∗ o At the relay, the symbols ṡ1 , . . . ṡn are modulated via an M -
+w(n+1)β+(i1 −1)β+i2 , (4)
PSK modulator, thus generating the phases ϕ̇1 , . . . , ϕ̇n , where
where Real {·} is the real part of the complex variable, ∗ is ϕ̇∂ ∈ (0, 2π] (1 6 ∂ 6 n). The transmitted symbols of the
conjugation operator and ϕ0 = 0. Based on Ad , the symbol d relay are conveyed by the reflected signals of the RIS. In
can be estimated as particular, the reflected signal of the l-th RIS reflecting element
( is expressed as
1, Ad > 0 
dest = . (5)
−1, Otherwise 
Ṫel,l = k, exp (j ϕ̇1 ) k, . . . , exp (j ϕ̇n ) k,
Then, the estimated symbol dest is converted into the estimated | {z }
bit best . Specifically, best = 1 when dest = 1, otherwise, best = Ṙd

0.
d˙ · k, d˙ · exp (j ϕ̇1 ) k, . . . , d˙ · exp (j ϕ̇n ) k
On the other hand, to perform the demodulation for the 
symbol s∂ (1 6 ∂ 6 n) of the source, the decision metric is | {z }
İ
first computed as h d
i
β   = Ṫ el,l,1 , . . . , Ṫ el,l,ℑ , . . . , Ṫ el,l,2(n+1)β , (8)
X
As,∂ = Zi∗ · Z∂β+i + Z(n+1)β+i
∗
· Z(n+1)β+∂β+i
i=1
where the signals Ṙd and İd can be treated as the reference
 !∗ part and information-bearing part in the transmission of sym-
β q Lsd
≈
X
 −ε
Dsd
X
Λsd,ρsd ki + wi bol d,˙ respectively. The symbol ḋ is conveyed by the state
i=1 ρsd =1
of signal İd . To be specific, the signal İd equals the positive
Lsd
! of signal Ṙd if d˙ = 1; otherwise (i.e., d˙ = −1), it equals the
q
−ε
X negative of signal Ṙd . In addition, the signals k (resp. ḋ·k) and
× Dsd Λsd,ρsd exp (jϕ∂ ) ki + w∂β+i
exp (j ϕ̇∂ ) k (resp. d˙ · exp (j ϕ̇∂ ) k) in Eq. (8) can be treated as
ρsd =1
!∗ reference part and information-bearing part in the transmission
q Lsd
−ε
X of symbol ṡ∂ (1 6 ∂ 6 n), respectively. The spreading factor
+ d2 Dsd Λsd,ρsd ki + w(n+1)β+i
SF of scheme II is defined as 2 (n + 1) β. Accordingly, the
ρsd =1 Ξ
q Lsd
!# data rate of the source and relay are RII,s = SFII,s ·Tc and
X ΞII,r
−ε
× Dsd Λsd,ρsd exp (jϕ∂ ) ki +w(n+1)β+∂β+i . RII,r = SF ·Tc , respectively.
ρsd =1 At the destination, the received baseband signal can be
(6) expressed as
Hence, based on As,∂ , the estimated symbol s∂,est can be q Lsr X
X Lrd X
N
−ε −ε
obtained by using M -PSK demodulation. Then the estimated ṙℑ = Dsr Drd αsr,l,ρsrαrd,l,ρrd Ṫ el,l,ℑ−τl,ρsr−τl,ρrd
bit sub-stream c∂,est can be obtained from an M -ary symbol- ρsr =1ρrd =1 l=1

to-bit converter. + wℑ , (9)

B. RIS-M-FM-DCSK System: Scheme II
Fig. 1(b) shows the system model for the RIS-M -FM-DCSK
with scheme II. In the CS-ii, the source communicates with
the destination through the RIS, which is treated as a relay. As
shown in Fig. 1(b), the information bit of the source is ḃ and
information bit stream of the relay is ċ = [ċ1 , . . . , ċ∂ , . . . , ċn ],
where the dimension of ċ∂ is 1 × log2 M . Then, the bit b is
converted into a symbol d˙ = 2ḃ − 1, while the bit sub-streams
ċ1 , . . . , ċn are converted to symbols ṡ1 , . . . ṡn by using an
M -ary bit-to-symbol converter. In each transmission period of
scheme II, the number of the information bits sent from the
where Dsr and Drd are the distances between the source and
relay and relay and destination, respectively; Lsr and Lrd are
the numbers of the paths between the source and l-th RIS
reflecting element and the l-th RIS reflecting element and
destination, respectively; αsr,l,ρsr and τl,ρsr denote channel
coefficient and delay of the ρsr -th (1 ≤ ρsr ≤ Lsr ) path
between the source and the l-th RIS reflecting element, re-
spectively; αrd,l,ρrd and τl,ρrd denote channel coefficient and
delay of the ρrd -th (1 ≤ ρrd ≤ Lrd ) path between the l-
th RIS reflecting element and the destination, respectively.
For convenience, a cascaded source-relay (RIS)-destination
channel via the l-th RIS reflecting element is modeled here.
 5

We define the number of the paths for the source-l-th RIS To detect the symbol s∂ (1 6 ∂ 6 n) of the source, the
reflecting element channel and the l-th RIS reflecting element- decision metric is first calculated as
destination channel as Lsr and Lrd , respectively. Thereby, the
β 
cascaded source-RIS (relay)-destination channel via the l-th X 
RIS reflecting element can be modeled as a Lsr Lrd -path fading Ȧs,∂ = Żi∗ · Ż ∂β+i + Ż(n+1)β+i
∗
· Ż (n+1)β+∂β+i
i=1
channel. The ρca -th (1 6 ρca 6 Lsr Lrd ) path with the channel  !∗
β L
coefficient Γsd,l,ρca is constructed by concatenating the ρsd -th X q
−ε −ε
X tot

path between the source and the l-th RIS reflecting element ≈  Dsr Drd Γsd,ρca ki + wi
i=1 ρca =1
and the ρsr -th path between the l-th RIS reflecting element and !
destination. The time delay of ρca -th path at the destination q L
X tot
−ε −ε
is τl,ρca = τl,ρsr + τl,ρrd . In this paper, we consider that the × Dsr Drd Γsd,ρca exp (j ϕ̇∂ ) ki + w∂β+i
path delays of each RIS reflecting element are the same, i.e., ρca =1
L
!∗
τl1 ,ρca = τl2 ,ρca (l1 = 1, . . . , N , l2 = 1, . . . , N , and l1 6= l2 ). q X tot
−ε −ε
Hence, τl,ρca can be simplified to τρca . + d2 Dsr Drd Γsd,ρca ki + w(n+1)β+i
ρca =1
Based on the above foundations, the received baseband !#
q L tot
signal can be rewritten as −ε −ε
X
× Dsr Drd Γsd,ρca exp (j ϕ̇∂ ) ki+w(n+1)β+∂β+i .
s ρca =1
L tot
1 X (14)
ṙℑ = ε Dε
Γsd,ρca Ṫ el,1,ℑ−τl,ρca + wℑ , (10)
Dsr rd ρ =1 ca
Hence, based on Ȧs,∂ , the estimated symbol ṡ∂,est can be
LP
obtained by using M -PSK demodulation. Then the estimated
tot
where Γsd,ρca is calculated as bit sub-stream ċ∂,est can be obtained from an M -ary symbol-
ρca =1 to-bit converter.

L
X tot Lsr X
X Lrd X
N
Γsd,ρca = αsr,l,ρsr αrd,l,ρrd
C. Channel Model
ρca =1 ρsr =1 ρrd =1 l=1
L
X tot N
X In this paper, we consider a multipath Rayleigh fading
= Γsd,l,ρca . (11) channel, which is commonly utilized in spread-spectrum
ρca =1 l=1 chaotic communication systems [21], [22], [25], [26].
In addition, we assume that the channel coefficients
Then,h the received basebandi signal ṙℑ is stored in a matrix αsd,l1 ,ρsd , αsr,l1 ,ρsr , and αrd,l1 ,ρrd are independent of
Ż = Ż 1 , Ż 2 , . . . , Ż 2(n+1)β . To detect the symbol d˙ of the αsd,l2 ,ρsd , αsr,l2 ,ρsr , and αrd,l2 ,ρrd , respectively [36], [38].
source, the decision metric is first computed as According to the central limit theorem [38], [39], the
random variables Λsd,ρsd and Γsd,ρca nfollow theo complex
  2
(n+1)β
X  Gaussian distributions CN 0, N E |αsd,l,ρsd | and
∗  n o n o
Ȧd= Real Ż i · Ż(n+1)β+i 2
CN 0, N E |αsr,l,ρsr | · E |αrd,l,ρrd |
2
, respectively,
 
i=1
(n+1 β q where E {·} denotes the expectation operation. In the next
L tot
XX
−ε −ε
X
section, the channel model and it corresponding distribution
≈ Real Dsr Drd Γsd,ρca exp j ϕ̇i1 −1 ki2
feature will be used for the performance analysis.
i1 =1 i2 =1 ρca =1
q L
X tot



˙ −ε −ε
+w(i1−1)β+i2 × d Dsr Drd Γsd,ρca exp j ϕ̇i1 −1 ki2
ρca =1
III. BER P ERFORMANCE A NALYSIS

∗ o
+w(n+1)β+(i1 −1)β+i2 , (12) In this section, the theoretical BER expressions of scheme I
and scheme II in the proposed RIS-M -FM-DCSK system are
where ϕ̇0 = 0. Based on Ȧd , the symbol d˙ can be estimated derived over a multipath Rayleigh fading channel. Without
as loss of generality, we make two assumptions here. First, we
assume
n that othe channel
n gains are o unequal [8],n[9], [45], i.e.,o
( 2 2 2
1, Ȧd > 0 E |αsd,l,1 | 6= E |αsd,l,2 | 6= · · · 6= E |αsd,l,ρsd | ,
d˙est =
n o n o n o
, (13)
−1, Otherwise E |αsr,l,1 |2 6= E |αsr,l,2 |2 6= · · · 6= E |αsr,l,ρsr |2 , and
n o n o n o
2 2 2
E |αrd,l,1 | 6= E |αrd,l,2 | 6= · · · 6= E |αrd,l,ρrd | .
Then, the estimated symbol d˙est is converted into the estimated Second, we assume that the symbol duration is much longer
bit ḃest . To be specific, ḃest = 1 when d˙est = 1, otherwise, than the path delays, i.e., 0 ≤ τρsd /τρca ≪ SF , thus the inter-
ḃest = 0. symbol interference can be ignored [25], [26], [45].
6

( n+1 β " Lsd q Lsd

)
XX X 2 2
X
2
−ε −ε ∗
Ad ≈ Real Dsd |Λsd,ρsd | |ϑi1 −1 | |ki2 | + Dsd Λsd,ρsd ϑi1 −1 ki2 w(n+1)β+(i1 −1)β+i2
i1 =1 i2 =1 ρsd =1 ρsd =1
n+1 β
" Lsd q Lsd
X X X X
= −ε
Dsd |Λsd,ρsd |2 |ϑi1 −1 |2 · |ki2 |2 + Dsd
−ε
[(Λsd,ρsd ,1 ki2 ,1 ϑi1 −1,1 − Λsd,ρsd ,1 ki2 ,2 ϑi1 −1,2
i1 =1 i2 =1 ρsd =1 ρsd =1
−Λsd,ρsd ,2 ki2 ,1 ϑi1 −1,2 − Λsd,ρsd ,2 ki2 ,2 ϑi1 −1,1 ) w(n+1)β+(i1 −1)β+i2 ,1 + (Λsd,ρsd ,1 ki2 ,1 ϑi1 −1,2 + Λsd,ρsd ,1 ki2 ,2 ϑi1 −1,1
Lsd
 q −ε X
+Λsd,ρsd ,2 ki2 ,1 ϑi1−1,1 −Λsd,ρsd ,2 ki2 ,2 ϑi1 −1,2) w(n+1)β+(i1−1)β+i2 ,2 + Dsd [(Λsd,ρsd ,1 ki2 ,1 ϑi1−1,1−Λsd,ρsd ,1 ki2 ,2 ϑi1−1,2
ρsd =1
−Λsd,ρsd ,2 ki2 ,1 ϑi1 −1,2 − Λsd,ρsd ,2 ki2 ,2 ϑi1 −1,1 ) w(i1 −1)β+i2 ,1 + (Λsd,ρsd ,1 ki2 ,1 ϑi1 −1,2 + Λsd,ρsd ,1 ki2 ,2 ϑi1 −1,1
+Λsd,ρsd ,2 ki2 ,1 ϑi1 −1,1 − Λsd,ρsd ,2 ki2 ,2 ϑi1 −1,2 ) w(i1 −1)β+i2 ,2 + w(n+1)β+(i1 −1)β+i2 ,1 w(i1 −1)β+i2 ,1


+w(n+1)β+(i1 −1)β+i2 ,2 w(i1 −1)β+i2 ,2 . (16)

A. Scheme I where Var {·} denotes the variance operation. In addition,

−ε PLsd 2
Dsd ρsd =1 |Λsd,ρsd | Es
The overall BER PI is related to PI,b and PI,c , where we define rs = N0 as the instantaneous
PI,b and PI,c are the BERs of the information bits b and c, received SNR for scheme I. Because Ad follows a Gaussian
respectively. Therefore, the overall BER PI is given by distribution with mean E {Ad } and variance Var {Ad }, the
PI,b PI,c nlog2 M BER of the information bit b conditioned on rs can be
PI = + . (15) expressed as
ΞI,tot ΞI,tot !
Without loss of generality, we assume that the information bit 1 E {Ad }
PI,b|rs = erfc p
b equals 1 in the following analysis. 2 2Var {Ad }
 
1) Derivation of PI,b : For the derivation of PI,b , we further
1 1
express Eq. (4) as Eq. (16). In Eq. (16), we define Λsd,ρsd = = erfc  q , (19)
2
Λsd,ρsd ,1 + jΛsd,ρsd ,2 , exp (jϕi1 −1 ) = ϑi1 −1 = ϑi1 −1,1 + 2 rs + β (n + 1) rs−2
−1

jϑi1 −1,2 , ki2 = ki2 ,1 + jki2 ,2 , and wℑ = wℑ,1 + jwℑ,2

(1 ≤ ℑ ≤ 2 (n + 1) β), where ϑ2i1 −1,1 + ϑ2i1 −1,2 = 1, where erfc (·) is the complementary error function [46]. Next,
Pβ 2 Es the average BER of the information bit b over a multipath
i2 =1 |ki2 | = 2(n+1) and Es is the transmitted energy per
symbol. More specifically, the average transmitted energy per Rayleigh fading channel is given by
Z ∞
bit is defined as Eb = ΞE s
in scheme I. Referring to [21],
I,tot
PI,b = PI,b|rs f (rs ) drs , (20)
[22], [45], for a large β, the random variable Ad follows a 0
Gaussian distribution. Hence, the mean and variance of the
where f (rs ) is the probability density function (PDF) of rs .
decision metric Ad can be respectively calculated as
The expression of f (rs ) is written as
−ε P LP
n+1 sd
2 2 Lsd  
Dsd |Λsd,ρsd | |ϑi1 −1 | Es X χρsd rs
i1 =1 ρsd =1 f (rs ) = exp − , (21)
E {Ad } = (n + 1) · r
ρ =1 ρsd ,ave
rρsd ,ave
2 (n + 1) sd

L
P sd where
−ε 2
Dsd |Λsd,ρsd | Es
ρsd =1 Lsd
= , (17)
Y rρsd ,ave
2 χρsd = , (22)
rρsd ,ave − rς,ave
ς=1,ς6=ρsd
n o
P LP
n+1 sd D−ε Es
−ε
Dsd
2
|Λsd,ρsd | |ϑi1 −1 | Es N0
2 in which rρsd ,ave = E |Λsd,ρsd |2 · sd N0 is the average
i1 =1 ρsd =1 value of the instantaneous received SNR on the ρsd -th path.
Var {Ad } =
2 (n + 1) 2) Derivation of PI,c : To derive PI,c , Eq. (6) is further
n+1
P expressed as
2
β N02 |ϑi1 −1 |
i1 =1
+ As,∂ ≈ Bsignal exp (jϕ∂ ) + Bnoise , (23)
2
−ε
L
P sd
2 where Bsignal and Bnoise are respectively written as
Dsd |Λsd,ρsd | Es N0 + βN02 (n + 1)
ρsd =1 Lsd
X
= , −ε
Bsignal = 2Dsd
2 2
|Λsd,ρsd | |ki | , (24)
2
(18) ρsd =1
 7

Bnoise 1) Derivation of PII,ḃ : To derive PII,ḃ , Eq. (12) is further

we define Γsd,ρsd =
β
" q Lsd
! expressed as Eq. (32). In Eq. (32),
X X

−ε ∗ ∗ Γsd,ρca ,1 + jΓsd,ρca ,2 , exp j ϕ̇i1 −1 = ϑ̇i1 −1 = ϑ̇i1 −1,1 +
= Dsd Λsd,ρsd ki × w∂β+i + w(n+1)β+∂β+i
i=1 ρsd =1 j ϑ̇i1 −1,2 , where ϑ̇2i1 −1,1 + ϑ̇2i1 −1,2 = 1. Similar to the variable
!
q Lsd
X   Ad , the decision metric Ȧd follows the Gaussian distribution.
−ε
+ Dsd Λsd,ρsd exp (jϕ∂ ) ki × wi∗ + w(n+1)β+i
∗
Hence, the mean and variance are respectively calculated as
sd ρ =1
 i −ε −ε
P LP
n+1 tot
2
2
∗ ∗
+ wi w∂β+i + w(n+1)β+i w(n+1)β+∂β+i . (25) n o Dsr Drd |Γsd,ρca | ϑ̇i1 −1 Es
i1 ρca =1
E Ȧd = (n + 1) ·
2 (n + 1)
For the demodulation of the M -PSK symbol exp (jϕ∂ ),
LP
tot
Bsignal , and Bnoise can be regarded as the desired component −ε −ε
Dsr Drd
2
|Γsd,ρca | Es
and noise component, respectively. According to [41], the ρca =1
= , (33)
equivalent conditional received SNR of the M -PSK symbol 2
exp (jϕ∂ ) is given by
−ε −ε
P LP
n+1 tot
2
2
E2 {Bsignal } n o Dsr Drd i =1 ρ =1 |Γsd,ρca | ϑ̇i1 −1 Es N0
rs,equ|rs = , (26) Var Ȧd =
1 ca
Var {Bnoise } 2 (n + 1)
n+1 2
where E {Bsignal } and Var {Bnoise } can be respectively cal- β
P 2
N0 ϑ̇i1 −1
culated as i =1
+ 1
L
P sd
2
2
−ε
Dsd |Λsd,ρsd | Es −ε −ε
LPtot
2
ρsd =1 Dsr Drd |Γsd,ρca | Es N0 + βN02 (n + 1)
E {Bsignal } = , (27) ρca =1
n+1 = .
2
(34)
L
P sd
−ε 2 LP
2Dsd |Λsd,ρsd | Es N0 Dsr Drd
−ε −ε
tot
|Γsd,ρca |2 Es
ρsd =1
2βN02 .
ρca =1
Var {Bnoise } = + (28) Also, we define ṙs = N0 as the instanta-
n+1 neous received SNR for scheme II. In addition, the relationship
Substituting Eqs. (27) and (28) into Eq. (26), one can obtain between the transmitted energy per symbol Es and average
transmitted energy per bit Eb is Es = (ΞII,s + ΞII,r ) Eb in
1
rs,equ|rs = 2 . (29) scheme II. Since the random
n variable
o Ȧd follows then Gaussian
o
2 (n + 1) rs−1 + 2β(n + 1) rs−2 distribution with mean E Ȧd and variance Var Ȧd , the
Afterwards, according to [41], the conditional BER of the conditional BER of the information bit ḃ can be expressed as
information bits c is given by  n o 
(
Q
p
2rs,equ|r


, M =2 1  E Ȧd 
 q s
 PII,ḃ|ṙs =erfc 
r

PI,c|rs = 2 2 π

, 2 n o
log2 M Q 2sin M rs,equ|rs , M ≥ 4 2Var Ȧd
(30)  
1 1
where Q (·) is the Gaussian Q-function [46]. Next, combining = erfc  q . (35)
2
Eqs. (21) and (30), the average BER of the information bits c 2 ṙs + β (n + 1) ṙs−2
−1

over a multipath Rayleigh fading channel is given by Because ṙs and rs follow the same distribution, the PDF of
Z ∞
ṙs can be attained by employing Eq. (21). Next, the average
PI,c = PI,c|rs f (rs ) drs . (31) BER for the transmitted bits of the source over a multipath
0
Rayleigh fading channel is given by
Finally, substituting Eqs. (20) and (31) into Eq. (15), one can Z ∞
obtain the overall BER of scheme I over a multipath Rayleigh PII,ḃ = PII,ḃ|ṙs f (ṙs ) dṙs . (36)
fading channel. 0
2) Derivation of PII,ċ : To derive the BER of the transmitted
bits from the relay to the destination PII,ċ , Eq. (14) is further
B. Scheme II expressed as
In scheme II, one information bit ḃ and nlog2 M information Ȧs,∂ ≈ Ḃsignal exp (j ϕ̇∂ ) + Ḃnoise , (37)
bits ċ are sent from the source and relay to the destination,
respectively, in each transmission period. Here, we define the where Ḃsignal and Ḃnoise are respectively calculated as
BERs of the information bits ḃ and ċ as PII,ḃ and PII,ċ , L
X tot
−ε −ε 2 2
respectively. Without loss of generality, we assume that the Ḃsignal = 2Dsr Drd |Γsd,ρca | |ki | , (38)
information bit ḃ equals 1 in the following analysis. ρca =1
8

( n+1 β " L q L
tot tot
X X X 2 X
−ε −ε 2 2 −ε −ε ∗
Ȧd ≈ Real Dsr Drd |Γsd,ρca | ϑ̇i1 −1 |ki2 | + Dsr Drd Γsd,ρca ϑ̇i1 −1 ki2 w(n+1)β+(i1 −1)β+i2
i1 =1 i2 =1 ρca =1 ρca =1
q Lsd
#)
X
−ε −ε
+ Dsr Drd ∗
Γ∗sd,ρca ϑ̇∗i1 −1 ki∗2 w(i1 −1)β+i2 +w(i1 −1)β+i2 w(n+1)β+(i1 −1)β+i2
ρca =1
n+1 β
" L q L
tot tot h
X X X 2 X
−ε −ε 2 2 −ε −ε
= Dsr Drd |Γsd,ρca | ϑ̇i1 −1 · |ki2 | + Dsr Drd Γsd,ρca ,1 ki2 ,1 ϑ̇i1 −1,1 − Γsd,ρca ,1 ki2 ,2 ϑ̇i1 −1,2
i1 =1 i2 =1 ρca =1 ca ρ =1
 
−Γsd,ρca ,2 ki2 ,1 ϑ̇i1−1,2−Γsd,ρca ,2 ki2 ,2 ϑ̇i1 −1,1 w(n+1)β+(i1−1)β+i2 ,1+ Γsd,ρca ,1ki2 ,1 ϑ̇i1−1,2+Γsd,ρca ,1ki2 ,2 ϑ̇i1−1,1+Γsd,ρca ,2 ki2 ,1 ϑ̇i1−1,1
 i q L
X tot h
−ε −ε
−Γsd,ρca ,2 ki2 ,2 ϑ̇i1−1,2 w(n+1)β+(i1 −1)β+i2 ,2 + Dsr Drd Γsd,ρca ,1 ki2 ,1 ϑ̇i1 −1,1 − Γsd,ρca ,1 ki2 ,2 ϑ̇i1 −1,2
ρ =1
  sd
−Γsd,ρca ,2 ki2 ,1 ϑ̇i1−1,2 −Γsd,ρca ,2 ki2 ,2 ϑ̇i1−1,1 w(i1−1)β+i2 ,1 + Γsd,ρca ,1 ki2 ,1 ϑ̇i1−1,2 +Γsd,ρca ,1 ki2 ,2 ϑ̇i1−1,1 +Γsd,ρca ,2 ki2 ,1 ϑ̇i1−1,1
 i
i
−Γsd,ρca ,2 ki2 ,2 ϑ̇i1 −1,2 w(i1−1)β+i2 ,2 + w(n+1)β+(i1 −1)β+i2 ,1 w(i1 −1)β+i2 ,1 + w(n+1)β+(i1 −1)β+i2 ,2 w(i1 −1)β+i2 ,2 . (32)

Ḃ noise Then, according to [41], the conditional BER of the informa-

β
" q L
! tion bits ċ is given by
X X tot
−ε −ε


= Dsr Drd Γ∗sd,ρca ki∗ × w∂β+i +w(n+1)β+∂β+i (
Q
p
2ṙs,equ|


q ṙs , M =2
i=1 ρca =1 PII,ċ|ṙs = 
 .
2 2 π
β q L tot
!
  log2 M Q 2sin M ṙ s,equ|ṙs , M ≥ 4
X X
−ε −ε
+ Dsr Drd Γsd,ρca exp (j ϕ̇∂ ) ki × wi∗ +w(n+1)β+i
∗
(44)
i=1 ρca =1
# Finally, the average BER for the transmitted bits for the relay
β  
X over a multipath Rayleigh fading channel is formulated as
+ wi∗ w∂β+i + w(n+1)β+i
∗
w(n+1)β+∂β+i . (39) Z ∞
i=1
PII,ċ = PII,ċ|ṙs f (ṙs ) dṙs . (45)
0
The signals Ḃsignal and Ḃnoise are the desired component and
noise component, respectively, for the demodulation of the IV. S IMULATION R ESULTS AND D ISCUSSION
M -PSK symbol exp (j ϕ̇∂ ). Hence, the equivalent conditional In this section, we evaluate the performance of the proposed
received SNR of the M -PSK symbol exp (j ϕ̇∂ ) can be for- RIS-M -FM-DCSK system with scheme I and scheme II by
mulated as [41] simulations. In simulations, we consider the following param-
n o eter setting for the multipath fading channels. In the CS-i,
E2 Ḃ signal for the link RIS→destination, the parameters ofnthe multipath
ṙs,equ|ṙs = n o, (40) o
Var Ḃ noise Rayleigh fading channel are set to Lsd = 2, E |αsd,l,1 |2 =
n o
2 2
n o n o 3 , E |α sd,l,2 | = 31 , τ1 = 0, and τ2 = 3Tc . In the CS-
where E Ḃ signal and Var Ḃ noise are respectively ii, for the link source→relay, the parameters ofnthe multipath o
2
Rayleigh fading channel are set to Lsr = 3, E |αsr,l,1 | =
n o n o
LP 1 2 1 2
= 61 , τ1 = 0, τ2 = 2Tc ,
tot
−ε −ε 2 , E |α | = , E |α |
o Dsr Drd ρ =1 |Γsd,ρca | Es 2 sr,l,2 3 sr,l,3
n
ca and τ3 = 4Tc ; for the link relay→destination, the parameters
E Ḃ signal = , (41)
n+1 ofnthe multipath o Rayleigh n fadingochannel are set to Lrd = 2,
2 2
4
E |αrd,l,1 | = 7 , E |αrd,l,2 | = 37 , τ1 = 0, and τ2 = Tc .
LP
tot
Furthermore, the path loss coefficient ε is set to 2 in both CSs
−ε −ε 2
n o 2Dsr Drd ρ =1 |Γsd,ρca | Es N0 [41].
+ 2βN02 .
ca
Var Ḃ noise =
n+1
(42) A. Performance Evaluation of the Proposed RIS-M -FM-
DCSK system
Substituting Eqs. (41) and (42) into Eq. (40), one can obtain 1) Scheme I: Fig. 2 shows the BER results of scheme I
with different values of M and N over a multipath Rayleigh
1
ṙs,equ|ṙs = 2 . (43) channel, where n is set to 2, SF = 300, and Dsd = 16m. It
2 (n + 1) ṙs−1 + 2β(n + 1) ṙs−2 can be seen that the theoretical BER results agree very well
 9

10 0 M =4, N =32 (Sim.) 10 0

M =4, N =32 (Theo.)
M =16, N =32 (Sim.)
M =16, N =32 (Theo.)
M =4, N =128 (Sim.)
10 -1 M =4, N =128 (Theo.) 10 -1
M =16, N =128 (Sim.)
M =16, N =128 (Theo.)

10 -2 10 -2
BER

BER
10 -3 10 -3

n=2, Source (Sim.)

n=2, Source (Theo.)
n=2, Relay (Sim.)
10 -4 10 -4 n=2, Relay (Theo.)
n=4, Source (Sim.)
n=4, Source (Theo.)
n=4, Relay (Sim.)
n=4, Relay (Theo.)
10 -5 10 -5
0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40
Eb /N0 Eb /N0

Fig. 2. Theoretical and simulated BER performance of scheme I in the RIS- (a)
M -FM-DCSK system over a multipath Rayleigh fading channel, where SF =
10 0
300, n = 2, and Dsd = 16m.

10 0 SF =240, n=1 (Sim.)

SF =240, n=1 (Theo.) 10 -1
SF =240, n=3 (Sim.)
SF =240, n=3 (Theo.)
SF =480, n=1 (Sim.)
10 -1 SF =480, n=1 (Theo.)
SF =480, n=3 (Sim.)
SF =480, n=3 (Theo.)
BER 10 -2

10 -2
10 -3
BER

n=2, Source (Sim.)

n=2, Source (Theo.)
10 -3
n=2, Relay (Sim.)
10 -4 n=2, Relay (Theo.)
n=4, Source (Sim.)
n=4, Source (Theo.)
n=4, Relay (Sim.)
10 -4 n=4, Relay (Theo.)
10 -5
0 5 10 15 20 25 30 35 40
Eb /N0
10 -5
0 5 10 15 20 25 30 35 40 (b)
Eb /N0
Fig. 4. Theoretical and simulated BER performance of scheme II in the RIS-
Fig. 3. Theoretical and simulated BER performance of scheme I in the RIS- M -FM-DCSK system over a multipath Rayleigh fading channel, where (a)
M -FM-DCSK system over a multipath Rayleigh fading channel, where SF = N = 64, (b) N = 128, SF = 480, M = 4, Dsr = 6m, and Drd = 10m.
300, M = 4, N = 128, and Dsd = 16m.

with the simulated ones in the BER region of practical interest.
As shown in Fig. 2, the BER performance of scheme I is
deteriorated with an increasing M . For instance, when N =
128, scheme I with M = 4 has about 2.5 dB performance
gain over scheme I with M = 16 at a BER of 10−4 over
a multipath Rayleigh channel. Moreover, as expected, we can
observe that the BER performance of scheme I improves as N
increases. Specifically, when M = 16, the BER performance
gap between N = 32 and N = 128 is about 5 dB at a BER
of 10−4 .
Fig. 3 shows BER performance of scheme I in the proposed
RIS-M -FM-DCSK system with different n and SF over a
multipath Rayleigh channel. It can be seen that the simulated
BER results match well with the theoretical BER results. From
this figure, one can observe that the BER of scheme I as n
increases. The main reason is that, when n increases, less
energy of the reference-part signal is required to transmit an
information bit. More specifically, the transmitted energy of
the reference part within the signal is shared by 1 + nlog2 M
bits. Besides, the BER of scheme I becomes worse with the
increase of SF , due to the fact that severer noise appears in
the detection process as SF increases.
2) Scheme II: Fig. 4 plots the theoretical and simulated
BER curves for the source and relay in scheme II over a
multipath Rayleigh fading channel, where the parameters M ,
SF , Dsr , and Drd are set to 4, 480, 6m, and 10m, respectively.
It can be seen that the theoretical BER curves agree well with
the simulated ones regardless of N or n. Moreover, the BER
performance of both the source and the relay in scheme II
improves as n increases, which is similar to the observation
in scheme I. Furthermore, the BER performance of the source
is better than that of the relay. The main reason is that the
Euclidean distance of the source is larger than that of the relay
in the signal space. For instance, when N = 128, the source
has about 3.5 dB performance gain over the relay at a BER
of 10−4 over a multipath Rayleigh fading channel.
In Fig. 5, the BER curves for the source and relay are plotted
against the parameter µsr = DsrD+D sr
rd
, where total distance for
10

10 0 10 0

10 -1 10 -1

10 -2 10 -2
BER

BER
10 -3 10 -3
Eb /N0 =20 dB 2
Eb /N0 =24 dB RIS-M -FM-DCSK, M =2, N =32, n=1, R =
SF · Tc
Eb /N0 =28 dB 2
RIS-M -FM-DCSK, M =2, N =64, n=1, R =
SF · Tc
10 -4 10 -4 OMC-MIMO-DCSK, Nt =2, Nr =1, R =
1
SF · Tc
1
OMC-MIMO-DCSK, Nt =4, Nr =1, R =
SF · Tc
1
FM-DCSK, R =
SF · Tc
10 -5 10 -5
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 5 10 15 20 25 30 35 40 45 50
µsr Eb /N0

(a) Fig. 6. BER performance of scheme I in the RIS-M -FM-DCSK system,

the OMC-MIMO-DCSK system, and the FM-DCSK system over a multipath
10 0
Rayleigh fading channel, where SF = 360 and Dsd = 16 m.

10 0

10 -1

10 -1
BER

10 -2
10 -2
BER

Eb /N0 =20 dB
Eb /N0 =24 dB

10 -3 Eb /N0 =28 dB 10 -3

10 -4
RIS-M -FM-DCSK,M =4,N =32,n=4
10 -4
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 RIS-M -FM-DCSK,M =4,N =64,n=4
µsr MISO-CIM-MC-M DCSK, Nca =2, Nt =2, M =16
MISO-CIM-MC-M DCSK, Nca =2, Nt =4, M =16
(b) 10 -5
0 5 10 15 20 25 30 35 40 45 50
Eb /N0
Fig. 5. BER performance versus the parameter µsr for the (a) source and (b)
relay of scheme II in the RIS-M -FM-DCSK system over a multipath Rayleigh Fig. 7. BER performance of scheme I in the RIS-M -FM-DCSK system
fading channel, where µsr = D D+D sr
, Dsr + Drd = 16m, N = 64, and the MISO-CIM-MC-M DCSK system over a multipath Rayleigh fading
sr rd
SF = 480, M = 4, and n = 2. channel, where SF = 360 and Dsd = 16m.

the link source →relay→destination (i.e., Dsr + Drd ) is set to

16m. We can find that when the relay is located at the position
that has the same distance to the source and destination, both
the source and relay have the worst BER performance. The
BER performance of both the source and relay gets better
as the relay moves closer to the source or destination. The
reason of this observation is related to the received SNR at the
destination. Specifically, for a fixed Eb /N0 , when µsr = 0.5,
the destination has the smallest received SNR. Therefore, in
scheme II, the relay should be deployed close to the source or
destination to achieve desirable transmission reliability.
B. BER Performance Comparison between the Proposed RIS-
M -FM-DCSK System and Existing DCSK Counterparts
1) Scheme I: To better illustrate the superiority of the
proposed RIS-M -FM-DCSK system, we compare the BER
performance of the proposed system with other existing DCSK
counterparts. Fig. 6 shows the BER performance of scheme
I, the FM-DCSK [15], and the OMC-MIMO-DCSK system
[28] over a multipath Rayleigh fading channel, where SF and
Dsd are set to 360 and 16 m, respectively. As seen, the data
rates R are set to SF2·Tc , SF1·Tc , and SF1·Tc for the proposed
scheme I (M/n = 2/1), the OMC-MIMO-DCSK, and the
FM-DCSK systems, respectively. Note that R = SF2·Tc is the
lowest achievable data rate for scheme I, whereas the OMC-
MIMO-DCSK and FM-DCSK systems have a fixed data rate
R = SF1·Tc . For the OMC-MIMO-DCSK system, the number
of transmit antennas Nt is set to 2 and 4, and the number
of receive antennas Nr is set to 1. It can be seen that the
proposed scheme I outperforms the OMC-MIMO-DCSK and
FM-DCSK systems over a multipath Rayleigh fading channel.
For example, the proposed scheme I with M/n/N = 2/1/32
has about 10 dB and 7 dB performance gains over the OMC-
MIMO-DCSK system with Nt /Nr = 2/1 and Nt /Nr = 4/1,
respectively, at a BER of 10−4 over a multipath Rayleigh
fading channel.
Fig. 7 depicts the BER results of the proposed scheme I and
the MISO-CIM-MC-M DCSK system [29] over a multipath
 11

10 0 Moreover, the buffer sizes for the protocol 1 and protocol 2

in the buffer-aided DCSK-SWIPT relay system are set to 10.
10 -1 In Fig. 8, the data rate R for the source and relay in scheme
II (M/n = 2/1) are set to R = SF1·Tc . However, the data rate
of the buffer-aided DCSK-SWIPT relay system is adaptive
10 -2
according to the channel condition, and its maximum achiev-
able rate is R = SF1·Tc . When Dsr = 2m and Drd = 6m,
BER

10 -3
the BER results for the proposed scheme II and the buffer-
aided DCSK-SWIPT relay system are shown in Fig. 8(a). It
10 -4 is obvious that the BER performance of both the source and
relay in scheme II is much better than that of the buffer-aided
10 -5 RIS-M -FM-DCSK, M =2,N =16, n=1, Source
DCSK-SWIPT relay system. For instance, both the source and
RIS-M -FM-DCSK, M =2, N =16, n=1, Relay
Buffer-aided DCSK-SWIPT relay system, Protocol 1
relay in scheme II with M/n = 2/1 have about 30 dB and
Buffer-aided DCSK-SWIPT relay system, Protocol 2 37 dB performance gains over the protocol 1 and protocol 2
10 -6
0 10 20 30 40 50 60 70 in the buffer-aided DCSK-SWIPT relay system, respectively,
Eb /N0
at a BER of 10−4 over a multipath Rayleigh fading channel.
(a) Besides, one can observe that the source and relay in scheme
10 0 II with M/n = 2/1 have similar BER performance.
In the case of Dsr = 4m and Drd = 4m (i.e., the relay
10 -1 is located at the same distance from both the source and
destination), the BER performance of the proposed scheme
10 -2
II and the buffer-aided DCSK-SWIPT relay system over a
multipath Rayleigh fading channel is presented in Fig. 8(b).
Although the deployment location of the relay is the most
BER

10 -3
unfavorable for the proposed scheme II (see the discussion
for Fig. 5), the BER performance of the proposed scheme II
10 -4 is still much better than that of the buffer-aided DCSK-SWIPT
relay system. For instance, both the source and the relay in
10 -5
scheme II with M/n = 2/1 have about 30 dB and 31 dB
RIS-M -FM-DCSK, M =2,N =16, n=1, Source
RIS-M -FM-DCSK, M =2, N =16, n=1, Relay performance gains with respect to the protocol 1 and protocol
Buffer-aided DCSK-SWIPT relay system, Protocol 1
Buffer-aided DCSK-SWIPT relay system, Protocol 2 2 in the buffer-aided DCSK-SWIPT relay system, respectively,
10 -6
0 10 20 30 40 50 60 70 at a BER of 10−3 over a multipath Rayleigh fading channel.
Eb /N0

(b)
Fig. 8. BER performance of scheme II in the RIS-M -FM-DCSK system and
the buffer-aided DCSK-SWIPT relay system over a multipath Rayleigh fading
channel, where SF = 200, (a) Dsr = 2m, Drd = 6m, and (b) Dsr = 4m,
Drd = 4m.
Rayleigh fading channel, where SF and Dsd are set to 360
and 16m, respectively. For a fair comparison, the data rates
of the proposed scheme I (M/n = 4/4) and MISO-CIM-MC-
M DCSK system (Nca /M = 2/16) are set to R = SF9·Tc . As
observed, the proposed scheme I has better BER performance
than the MISO-CIM-MC-M DCSK system. For instance, the
proposed scheme I with M/n/N = 4/4/32 has about
5 dB performance gain with respect to the MISO-CIM-MC-
M DCSK system with Nca /M/Nt = 2/16/4 at a BER of
10−4 over a multipath Rayleigh fading channel.
Remark 1: Although the existing OMC-MIMO-DCSK and
MISO-CIM-MC-M DCSK systems could achieve better BER
performance by increasing the number of antennas, this would
significantly increase the cost and hardware complexity.
2) Scheme II: Considering the CS-ii, we compare the BER
performance between the proposed scheme II and buffer-aided
DCSK-SWIPT relay system [33] over a multipath Rayleigh
fading channel in Fig. 8. The relays in the proposed scheme
II and buffer-aided DCSK-SWIPT relay system are passive.
V. C ONCLUSION
In this paper, we have investigated the design and perfor-
mance analysis of a novel RIS-M -FM-DCSK system for real-
izing high-reliability transmissions in wireless environments.
To achieve this objective, we have conceived two efficient
transmission schemes for the proposed RIS-M -FM-DCSK
system, referred to as scheme I and scheme II. Specifically,
in scheme I, the RIS is treated as a transmitter at the
source. Moreover, the information bits are carried by both
the positive/negative state of the FM chaotic signal and the
M -PSK symbols, which are obtained by the phase adjustment
of the RIS. Thereby, scheme I is suitable for the CS that the
direct link exists between the source and the destination. On
the contrary, scheme II is suitable for the CS that the direct
link between the source and the destination is blocked by an
obstacle, such that the RIS is treated as a relay. In scheme II,
the information bits of the source and relay can be sent to the
destination simultaneously. In particular, the information bits
of the source and relay are carried by the positive/negative
state of the FM chaotic signal and the M -PSK symbols
generated by the RIS, respectively. Furthermore, the theoretical
BER expressions of the proposed RIS-M -FM-DCSK system
over multipath Rayleigh fading channels have been derived.
Simulated BER results match well with the corresponding
12
theoretical derivations, demonstrating the accuracy of the
proposed theoretical expressions. In addition, the proposed
RIS-M -FM-DCSK system significantly outperforms the FM-
DCSK, OMC-MIMO-DCSK, MISO-CIM-MC-M DCSK, and
buffer-aided DCSK-SWIPT relay systems in terms of BER.
As a consequence, the proposed RIS-M -FM-DCSK system
can be considered as a promising candidate for low-power
and low-cost wireless communication networks.
R EFERENCES
[1] F. C. M. Lau and C. K. Tse, Chaos-based Digital Communication
Systems. Springer, 2003.
[2] C. Bai, H. P. Ren, and G. Kolumbán, “Double-sub-stream M-ary
differential chaos shift keying wireless communication system using
chaotic shape-forming filter,” IEEE Trans. Circuits Syst. I, Reg. Papers,
vol. 67, no. 10, pp. 3574–3587, Oct. 2020.
[3] Z. Chen, L. Zhang, W. Wang, and Z. Wu, “A pre-coded multi-carrier
M-ary chaotic vector cyclic shift keying transceiver for reliable commu-
nications,” IEEE Trans. Wireless Commun., vol. 21, no. 2, pp. 1–1, Feb.
2022.
[4] Z. Zhou, T. Zhou, and J. Wang, “Performance of multiple-access DCSK
communication over a multipath fading channel with delay spread,”
Circuits Syst. Signal Process., vol. 27, no. 4, pp. 507–518, Aug. 2008.
[5] Y. Xia, C. K. Tse, and F. C. M. Lau, “Performance of differential chaos-
shift-keying digital communication systems over a multipath fading
channel with delay spread,” IEEE Trans. Circuits Syst. II, Express Briefs,
vol. 51, no. 12, pp. 680–684, Dec. 2004.
[6] M. H. L. Filipović and T. Matić, “Influence of human body motion on
DCSK performance over HBC channel,” in Proc. Int. Conf. Smart Syst.
Technol., Oct. 2018, pp. 159–162.
[7] M. Miao, L. Wang, G. Chen, and W. Xu, “Design and analysis of
replica piecewise M-Ary DCSK scheme for power line communications
with asynchronous impulsive noise,” IEEE Trans. Circuits Syst. I, Reg.
Papers, vol. 67, no. 12, pp. 5443–5453, Dec. 2020.
[8] M. Chen, W. Xu, D. Wang, and L. Wang, “Multi-carrier chaotic
communication scheme for underwater acoustic communications,” IET
Commun., vol. 12, no. 14, Aug. 2019.
[9] X. Cai, W. Xu, L. Wang, and G. Kaddoum, “Joint energy and correlation
detection assisted non-coherent OFDM-DCSK system for underwater
acoustic communications,” IEEE Trans. Commun., pp. 1–1, Apr. 2022.
[10] R. Luo, H. Yang, C. Meng, and X. Zhang, “A novel SR-DCSK-based
ambient backscatter communication system,” IEEE Trans. Circuits Syst.
II, Express Briefs, vol. 69, no. 3, pp. 1707–1711, Mar. 2022.
[11] Z. Chen, L. Zhang, J. Zhang, Z. Wu, and D. Luobu, “An OFDM-based
pre-coded chaos shift keying transceiver for reliable V2V transmission,”
IEEE Trans. Veh. Technol., pp. 1–1, 2022.
[12] L. Dai, Y. Fang, Z. Yang, P. Chen, and Y. Li, “Protograph LDPC-coded
BICM-ID with irregular CSK mapping in visible light communication
systems,” IEEE Trans. Veh. Technol., vol. 70, no. 10, pp. 11 033–11 038,
Oct. 2021.
[13] Y. Fang, Y. Bu, P. Chen, F. C. M. Lau, and S. A. Otaibi, “Irregular-
mapped protograph LDPC-coded modulation: a bandwidth-efficient so-
lution for 6G-enabled mobile networks,” IEEE Trans. Intell. Transp.
Syst., pp. 1–14, Nov. 2021.
[14] D. Song, L. Wang, Z. Xu, and G. Chen, “Joint code rate compatible
design of DP-LDPC code pairs for joint source channel coding over
implant-to-external channel,” IEEE Trans. Wireless Commun., pp. 1–1,
Jan. 2022.
[15] M. P. Kennedy, G. Kolumban, G. Kis, and Z. Jako, “Performance
evaluation of FM-DCSK modulation in multipath environments,” IEEE
Trans. Circuits Syst. I, Fundam. Theory, vol. 47, no. 12, pp. 1702–1711,
Dec. 2000.
[16] L. Wang, X. Min, and G. Chen, “Performance of SIMO FM-DCSK
UWB system based on chaotic pulse cluster signals,” IEEE Trans.
Circuits Syst. I, Reg. Papers, vol. 58, no. 9, pp. 2259–2268, Sept. 2011.
[17] S. Chen, L. Wang, and G. Chen, “Data-aided timing synchronization
for FM-DCSK UWB communication systems,” IEEE Transactions on
Industrial Electronics, vol. 57, no. 5, pp. 1538–1545, May. 2010.
[18] H. Ma, G. Cai, Y. Fang, J. Wen, P. Chen, and S. Akhtar, “A new
enhanced energy-detector-based FM-DCSK UWB system for tactile
internet,” IEEE Trans. Ind. Informat., vol. 15, no. 5, pp. 3028–3039,
May 2019.
[19] W. K. Xu, L. Wang, and G. Kolumbán, “A novel differential chaos shift
keying modulation scheme,” Int. J. Bifurcation Chaos, vol. 21, no. 03,
pp. 799–814, 2011.
[20] Y. Lu, M. Miao, L. Wang, and W. Xu, “A multilevel code-shifted
differential chaos shift keying system with reference diversity,” IEEE
Trans. Circuits Syst. II, Express Briefs, vol. 67, no. 11, pp. 2462–2466,
Nov. 2020.
[21] G. Kaddoum, “Design and performance analysis of a multiuser OFDM
based differential chaos shift keying communication system,” IEEE
Trans. Commun., vol. 64, no. 1, pp. 249–260, Jan 2016.
[22] Y. Tao, Y. Fang, H. Ma, S. Mumtaz, and M. Guizani, “Multi-carrier
DCSK with hybrid index modulation: a new perspective on frequency-
index-aided chaotic communication,” IEEE Trans. Commun., pp. 1–1,
Apr. 2022.
[23] W. Xu and L. Wang, “CIM-DCSK: A differential chaos shift keying
scheme with code-index modulation,” in Proc. Int. Symp. Commun. Inf.
Technol., 2016, pp. 100–104.
[24] W. Xu, Y. Tan, F. C. M. Lau, and G. Kolumbán, “Design and op-
timization of differential chaos shift keying scheme with code index
modulation,” IEEE Trans. Commun., vol. 66, no. 5, pp. 1970–1980,
May 2018.
[25] X. Cai, W. Xu, F. C. M. Lau, and L. Wang, “Joint carrier-code index
modulation aided M -ary differential chaos shift keying system,” IEEE
Trans. Veh. Technol., vol. 69, no. 12, pp. 15 486–15 499, Dec. 2020.
[26] X. Cai, W. Xu, L. Wang, and G. Chen, “Towards high-data-rate
noncoherent chaotic communication: a multiple-mode differential chaos
shift keying system,” IEEE Trans. Wireless Commun., vol. 20, no. 8, pp.
4888–4901, Aug. 2021.
[27] P. Chen, L. Wang, and F. C. M. Lau, “One analog STBC-DCSK
transmission scheme not requiring channel state information,” IEEE
Trans. Circuits Syst. I, Reg. Papers, vol. 60, no. 4, pp. 1027–1037, Apr.
2013.
[28] S. Wang, S. Lu, and E. Zhang, “MIMO-DCSK communication scheme
and its performance analysis over multipath fading channels,” J. Syst.
Eng. Electron., vol. 24, no. 5, pp. 729–733, Oct. 2013.
[29] G. Cai, Y. Fang, P. Chen, G. Han, G. Cai, and Y. Song, “Design of
an MISO-SWIPT-aided code-index modulated multi-carrier M -DCSK
system for e-health IoT,” IEEE J. Sel. Areas Commun., vol. 39, no. 2,
pp. 311–324, Feb. 2021.
[30] G. Kaddoum, F. Parzysz, and F. Shokraneh, “Low-complexity amplify-
and-forward relaying protocol for non-coherent chaos-based communi-
cation system,” IET Commun., vol. 8, pp. 2281–2289, Sept. 2014.
[31] G. Cai, Y. Fang, G. Han, J. Xu, and G. Chen, “Design and analysis
of relay-selection strategies for two-way relay network-coded DCSK
Systems,” IEEE Trans. Veh. Technol., vol. 67, no. 2, pp. 1258–1271,
Feb. 2018.
[32] Y. Zhang, Y. F. G. Cai, and P. Chen, “Performance of M -ary DCSK
modulation in SWIPT cooperative system with decoding cost,” in Proc.
Int. Symp. Med. Inf. Commun. Technol., Apr. 2021, pp. 227–232.
[33] M. Qian, G. Cai, Y. Fang, and G. Han, “Design of link-selection
strategies for buffer-aided DCSK-SWIPT relay system,” IEEE Trans.
Commun., vol. 68, no. 10, pp. 6023–6038, Oct. 2020.
[34] S. Basharat, S. A. Hassan, H. Pervaiz, A. Mahmood, Z. Ding, and
M. Gidlund, “Reconfigurable intelligent surfaces: potentials, applica-
tions, and challenges for 6G wireless networks,” IEEE Wireless Com-
mun., vol. 28, no. 6, pp. 184–191, Dec. 2021.
[35] G. C. Alexandropoulos, N. Shlezinger, and P. del Hougne, “Reconfig-
urable intelligent surfaces for rich scattering wireless communications:
recent experiments, challenges, and opportunities,” IEEE Commun.
Mag., vol. 59, no. 6, pp. 28–34, Jun. 2021.
[36] Q. Wu, S. Zhang, B. Zheng, C. You, and R. Zhang, “Intelligent re-
flecting surface-aided wireless communications: a tutorial,” IEEE Trans.
Commun., vol. 69, no. 5, pp. 3313–3351, May 2021.
[37] Q. Li, M. Wen, S. Wang, G. C. Alexandropoulos, and Y. C. Wu, “Space
shift keying with reconfigurable intelligent surfaces: phase configuration
designs and performance analysis,” IEEE Open J. Commun. Soc., vol. 2,
pp. 322–333, 2021.
[38] E. Basar, M. Di Renzo, J. De Rosny, M. Debbah, M. S. Alouini, and
R. Zhang, “Wireless communications through reconfigurable intelligent
surfaces,” IEEE Access, vol. 7, pp. 116 753–116 773, 2019.
[39] E. Basar, “Transmission through large intelligent surfaces: a new frontier
in wireless communications,” in Proc. Eur. Conf. Networks Commun.,
Jun. 2019, pp. 112–117.
[40] J. Yuan, M. Wen, Q. Li, E. Basar, G. C. Alexandropoulos, and G. Chen,
“Receive quadrature reflecting modulation for RIS-empowered wireless
communications,” IEEE Trans. Veh. Technol., vol. 70, no. 5, pp. 5121–
5125, May 2021.
 13

[41] J. G. Proakis and M. Salehi, Digital communications. New York, NY,

USA: McGraw-Hill, 2007.
[42] A. Kumar and P. R. Sahu, “Performance analysis of differential chaos
shift keying modulation with transmit antenna selection,” IET Commun.,
vol. 10, pp. 327–335, Feb. 2016.
[43] W. Hu, L. Wang, and G. Kaddoum, “Design and performance analysis of
a differentially spatial modulated chaos shift keying modulation system,”
IEEE Trans. Circuits Syst. II, Express Briefs, vol. 64, no. 11, pp. 1302–
1306, 2017.
[44] L. Wang, C. Zhang, and G. Chen, “Performance of an SIMO FM-DCSK
communication system,” IEEE Trans. Circuits Syst. II, Express Briefs,
vol. 55, no. 5, pp. 457–461, May 2008.
[45] Z. Chen, L. Zhang, and Z. Wu, “High data rate discrete-cosine-spreading
aided M -ary differential chaos shift keying scheme with low PAPR,”
IEEE Trans. Circuits Syst. II, Express Briefs, vol. 67, no. 11, pp. 2492–
2496, Nov. 2020.
[46] M. K. Simon and M. S. Alouini, Digital Communication over Fading
Channels, 2nd ed. New York, USA: Wiley, 2005.